Coastal ecosystems simulation: a decision tree analysis for Bivalve's growth conditions

Abstract

The usage of data mining models has the main purpose of discovering new patterns from dataset analysis by extracting knowledge from data and converting it to information. The most challenging part of problem solving is not the generation of high number of instances in dataset, most often hard to understand, but the interpretation of all those instances to extrapolate information about it. Simulation of coastal ecosystems is used to replicate some real conditions related with physical, chemical and biological processes, and produces large datasets from which it could be deduced some information about attributes behaviors. This paper relates the use of Decision Tree models to analyze the growth of bivalve species in an ecosystem simulation. With a set of attributes that represents the water quality in certain modeled regions, the usage of Decision Tree is intended to identify the most significant attribute conditions, which could justify the growth behavior for each analyzed species. This approach aims the creation of new information about how water conditions should be to promote a healthy and fast growth of the analyzed species, being useful to know in which zones the bivalve should be seeded, and which are the conditions that aquaculture producers should afford to benefit the quality of its crops.